kf_config.json

{
  "library_name": "kerasformers",
  "kerasformers_version": "1.2.1",
  "model_module": "kerasformers.models.depth_anything_v2",
  "model_class": "DepthAnythingV2DepthEstimation",
  "variant": "depth_anything_v2_base",
  "weights": "model.weights.h5",
  "schema_version": 2,
  "weight_dtype": "float32",
  "model_type": "depth_anything",
  "vision_config": {
    "backbone_dim": 768,
    "backbone_depth": 12,
    "backbone_num_heads": 12,
    "out_indices": [
      3,
      6,
      9,
      12
    ],
    "neck_hidden_sizes": [
      96,
      192,
      384,
      768
    ],
    "fusion_hidden_size": 128,
    "reassemble_factors": [
      4,
      2,
      1,
      0.5
    ],
    "depth_estimation_type": "relative",
    "max_depth": 1.0,
    "image_size": 518
  }
}